Motor vehicle crashes are the major cause of death and injury among adolescents. Crash risk is particularly elevated at licensure, declines rapidly for about 6 months (consistent with an effect of learning), and then decline more slowely for a number of years. Risk is particularly elevated under certain driving conditions, such as with teen passengers, at night, while engaged in secondary in-vehicle tasks. Our group has developed a program of research, including observational and experimental studies, designed to increase understanding of teen driving risks and reduce crash risks during the early months of licensure. We are conducting naturalistic observational studies to determine the nature of teen driving risk. Specifically, we are interested in determining how driving performance improves over time and varies under certain driving conditions, such as with teen passengers and at night. In the Naturalistic Teenage Driving Study and the Supervised Practice Driving Study we instrumented the vehicles of teenage drivers with accelerometers, GPS, and cameras with continuous recording, enabling the assessment of virtually all aspects of driving over the first 18 months of driving. Because teens shared vehicles with their parents, who were also study participants, we could compare teen and adult driving behavior in the same vehicles and driving conditions. Findings to date have included high crash rates among teens compared to parents, declining over time; high rates of elevated g-force events among teens that did not decline over time. In the Teen Passenger Study we are examing the effect of peer passengers on the simulated risky driving of teenage male drivers. To date we have demonstrated that teens drive in a more risky manner in the presence of passengers, particularly risk accepting passengers. Teens sensive to exclusion were particularly sensitive to peer influence.